Audio LLM–EEG Similarity Alignment on Alice in Wonderland sentence-averaged
0.2857Pearson RSAAudio-Flamingo-3
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Audio-Flamingo-32026.01 | 0.2857 | 0.1748 | 0.1092 | 0.3342 | 0.2233 | 0.0193 | 0.2233 | 0.3112 | |
| GLM-4-Voiceparameters=9B2026.01 | 0.2597 | 0.2031 | 0.1376 | 0.3008 | 0.2011 | 0.0162 | 0.2011 | 0.2783 | |
| Granite-Speech-3.3parameters=8B2026.01 | 0.2531 | 0.1949 | 0.1387 | 0.3813 | 0.2226 | 0.0179 | 0.2226 | 0.3154 | |
| SpeechGPT-2.0-previewparameters=7B2026.01 | 0.2527 | 0.1931 | 0.1347 | 0.332 | 0.2027 | 0.0168 | 0.2027 | 0.2915 | |
| Llama-3.1-Omniparameters=8B2026.01 | 0.2396 | 0.1765 | 0.1144 | 0.458 | 0.2777 | 0.0268 | 0.2777 | 0.396 | |
| MiniCPM-oversion=2.62026.01 | 0.2376 | 0.1749 | 0.1229 | 0.3113 | 0.1999 | 0.0158 | 0.1999 | 0.2762 | |
| Qwen2-Audioparameters=7B, variant=Instruct2026.01 | 0.2368 | 0.1773 | 0.1334 | 0.3138 | 0.1931 | 0.0151 | 0.1931 | 0.2767 | |
| Qwen2-Audioparameters=7B2026.01 | 0.2306 | 0.1848 | 0.1248 | 0.2962 | 0.1788 | 0.0139 | 0.1788 | 0.254 | |
| Baichuan-Audiovariant=Instruct2026.01 | 0.2286 | 0.169 | 0.1125 | 0.4067 | 0.2394 | 0.0205 | 0.2394 | 0.3486 | |
| Ultravox-v0.5backbone=LLaMA-3.1-8B2026.01 | 0.2216 | 0.1377 | 0.0973 | 0.4722 | 0.2813 | 0.0243 | 0.2813 | 0.3993 | |
| Ultravox-v0.5backbone=LLaMA-3.2-1B2026.01 | 0.2209 | 0.1378 | 0.0936 | 0.4691 | 0.2751 | 0.0231 | 0.2751 | 0.394 | |
| Baichuan-Audiovariant=Base2026.01 | 0.2028 | 0.1527 | 0.1107 | 0.3774 | 0.2216 | 0.019 | 0.2216 | 0.3256 |